What is dynamic airflow

[SiriusC1024]

Oh I see what you're saying. The WOT pull itself is a transient (airflow wise), yet you want data there. During the pull TPS isn't transient but airflow is. No this filter gets rid of that. That's a limitation of this filter.

You're wanting to allow that data to pass with an OR statement. Like ...OR TPS>95?

[GHuggins]

You'll have to explain. I see every spike in TPS (pedal) being filtered. Are you asking how to differentiate between PE and normal?

It's too damn late for this for one thing - 3am , but yes it's filtering fantastically however that's the problem I need it more sensitive to where it will show the data still with the airflow changing a somewhat decent amount. This particular car should be making decently to the tires, so flowing some air.

Attaching a screenshot showing what I'm trying to explain.

[GHuggins]

Oh I see what you're saying. The WOT pull itself is a transient (airflow wise), yet you want data there. During the pull TPS isn't transient but airflow is. No this filter gets rid of that. That's a limitation, yes.

Any way around it?

[SiriusC1024]

I think so. Add OR TPS>95

ABS([50040.71]-[50040.71.shift(-10)]) < ([50040.71]/[50040.71.shift(-10)])*0.1 & ABS([2311.71]-[2311.71.shift(-10)]) < ([2311.71]/[2311.71.shift(-10)])*0.1 OR [50090.156] > 95

Looks like this:
filter_tps_mod.png

Is that what you wanted?

[GHuggins]

I think so. Add OR TPS>95

ABS([50040.71]-[50040.71.shift(-10)]) < ([50040.71]/[50040.71.shift(-10)])*0.1 & ABS([2311.71]-[2311.71.shift(-10)]) < ([2311.71]/[2311.71.shift(-10)])*0.1 OR [50090.156] > 95

Looks like this:
filter_tps_mod.png

Is that what you wanted?

Oh, that is much better. Any drawbacks?

Is there still a way to make it more sensitive in general to grab just a little of the transient - I assume stupid question and possibly impossible on this one?

[GHuggins]

This is what that last change got, so much better.

[SiriusC1024]

The drawback is that at OR TPS>95 the transient terms are re-introduced into the DynAir model. Depends on what you're trying to do.

Just comparing to 2% TPS slope filter, which this filter is supposed to replace, this filter is more accurate. With the addition of >95% pedal (WOT) it's more complete. Wherever you found use for 2% TPS slope filter you'd use this filter instead.

As far as transients I figured if accurate MAF and VVE was established where the transient occurs then the ECU can better predict for that transient. This filter should make the data collected there more accurate, thereby diminishing the transients after changes based on filter data are applied.

[GHuggins]

The drawback is that at OR TPS>95 the transient terms are re-introduced into the DynAir model. Depends on what you're trying to do.

Just comparing to 2% TPS slope filter, which this filter is supposed to replace, this filter is more accurate. With the addition of >95% pedal (WOT) it's more complete. Wherever you found use for 2% TPS slope filter you'd use this filter instead.

As far as transients I figured if accurate MAF and VVE was established where the transient occurs then the ECU can better predict for that transient. This filter should make the data collected there more accurate, thereby diminishing the transients after changes based on filter data are applied.

So what you're saying is technically once things are dialed in closer it should theoretically populate the table with more data as it's then registering somewhat steady state? Hmmm wonder if there's an actual setting in the cal to define a transient mass? Hmmm, down the road perhaps..... Thinking other aspects in the gen 5's at this point

Thank You Again!

[SiriusC1024]

Yes, so your job is to make it register somewhat steady state. This filter would be the indicator. Remember that more data accumulates with a more steady pedal. That might sound confusing since it's not paying attention to pedal movement. The key is you have to give time for airflow to stabilize vs a given pedal movement. A quick stab is too short of a time. Since the window of stability is 2*sample interval (20ms for k=10ms, or 50 times per second), that means that pedal can be modulated quite a bit more than 2%. Try a longer span of pedal movements. I'd like to see if this can do what you need it to do.

Thanks for the feedback.

[SiriusC1024]

Here's what I mean. This is a clip from your log, GHuggins.
filter comparison.png

Comparison of Modified filter with your TPS>95 to account for WOT (WOT TRUE clearly seen at two locations)...

ABS([50040.71]-[50040.71.shift(-10)]) < ([50040.71]/[50040.71.shift(-10)])*0.1 & ABS([2311.71]-[2311.71.shift(-10)]) < ([2311.71]/[2311.71.shift(-10)])*0.1 OR [50090.156] > 95

...to a TPS filter (which I assume is a typical TPS filter format):

(abs([50090.156.slope(1500)])+abs([50090.156.slope(-500)]))<2

Modified filter allows for not just more data but more varied data to be collected. It also rejects some points of what would be considered stable in the TPS filter where airflow is in fact not steady state.

[TheMechanic]

So real world use is Gen 4 only, and this filter is a math function for a graph or a chart? Do you have it already saved as a math, chart, or graph? Mind sharing?

[Cringer]

So real world use is Gen 4 only, and this filter is a math function for a graph or a chart? Do you have it already saved as a math, chart, or graph? Mind sharing?

The dynamic air formulas work on Gen5's too. However, I believe the filtering that SiriusC is working on should be MUCH easier because Gen5's have the Flow Factor PID [2376] that tells numerically how much of the input is being scaled from MAF or VVE...basically we don't have to try to calculate or emulate the internal algorithm from the ECM for steady state or transient.

So you can then choose to filter out everything the ECM deems as not steady state (like we are doing here for Gen4)...or build that into the formula itself to back calculate MAF and VVE as GRG already did:
https://www.youtube.com/watch?v=WoXTmdTQkpE
Note, I have not tried this on a Gen5, so I cannot comment on how well I like it or how well it works.

[hjtrbo]

whilst you guys are figuring the filters out, i thought i'd have a crack at the zone edit tool. cringer, i'll hand the source to you when i'm done.

Screenshot 2023-08-24 235514.png

[GHuggins]

Here's what I mean. This is a clip from your log, GHuggins.
filter comparison.png

Comparison of Modified filter with your TPS>95 to account for WOT (WOT TRUE clearly seen at two locations)...

ABS([50040.71]-[50040.71.shift(-10)]) < ([50040.71]/[50040.71.shift(-10)])*0.1 & ABS([2311.71]-[2311.71.shift(-10)]) < ([2311.71]/[2311.71.shift(-10)])*0.1 OR [50090.156] > 95

...to a TPS filter (which I assume is a typical TPS filter format):

(abs([50090.156.slope(1500)])+abs([50090.156.slope(-500)]))<2

Modified filter allows for not just more data but more varied data to be collected. It also rejects some points of what would be considered stable in the TPS filter where airflow is in fact not steady state.

So if one wanted they could use this second older style filter that would go by tps alone then? I'll give it a try when I get a chance. Have several things to finish up right now. Probably tonight again.

I purposely used that higher air flowing ZL1's log to test the limits of the filter and the equation on this 5th gen. It should be making somewhere between 7 and 8 hundred to the tires in it's current state and a touch more on E. Seeing this one pass the test I think we'll be OK.

I also use this to back calculate the MAF on SD builds as the MAF input into the ECM is still very important even with it being failed since it's still used for torque and airmass calcs. Think this is the right one... 100 * ([50040.71] - [2311.71]) / [2311.71] + ([50114.156] + [50116.156]) Has nothing to do with any of this I guess, but if anyone wants it they can use it.




Nathan - VE tool mods are looking good too

[GHuggins]

So real world use is Gen 4 only, and this filter is a math function for a graph or a chart? Do you have it already saved as a math, chart, or graph? Mind sharing?

Sirius has them posted - a little confusing, but they're the "code's" he's been posting. The corrected filters are above and the formula's are a page back. Work on 4th and 5th gens as far as I can tell.

Guess I could post some other logs to use where I know the VVE is dialed in much much better or at least check them myself for accuracy, but I believe that's already been done and that zl1 log where I'm still fine tuning density for fuel delivery at different temps and what not should be a good example of this tools use since you can't fail the MAF on those cars AND I had already use the cylinder airmass calc to get those numbers, so should illustrate the difference in the two calcs too.

[SiriusC1024]

So real world use is Gen 4 only, and this filter is a math function for a graph or a chart? Do you have it already saved as a math, chart, or graph? Mind sharing?

It's a filter to use with a chart or graph. I don't see any reason it wouldn't work on gen 3-gen 5.

In fact, I was mentioning using the UMAP component to filter transients on my gen 3 running in SD.

For that it would be

ABS([2311.71]-[2311.71.shift(-10)]) < ([2311.71]/[2311.71.shift(-10)])*0.1 OR [50090.156] > 95

[SiriusC1024]

The dynamic air formulas work on Gen5's too. However, I believe the filtering that SiriusC is working on should be MUCH easier because Gen5's have the Flow Factor PID [2376] that tells numerically how much of the input is being scaled from MAF or VVE...basically we don't have to try to calculate or emulate the internal algorithm from the ECM for steady state or transient.

So you can then choose to filter out everything the ECM deems as not steady state (like we are doing here for Gen4)...or build that into the formula itself to back calculate MAF and VVE as GRG already did:
https://www.youtube.com/watch?v=WoXTmdTQkpE
Note, I have not tried this on a Gen5, so I cannot comment on how well I like it or how well it works.

That's cool. The more I was thinking about how to make a filter for the predicted dynamic airflow I began to realize it couldn't be a filter. The filter isn't meant to call a function, which is what it would take. There are multiple tables referenced, and writing that as filter would be just nope.

[SiriusC1024]

I purposely used that higher air flowing ZL1's log to test the limits of the filter and the equation on this 5th gen. It should be making somewhere between 7 and 8 hundred to the tires in it's current state and a touch more on E. Seeing this one pass the test I think we'll be OK.

That TPS filter is from Cringer's set of filters for DynAir tuning. https://forum.hptuners.com/showthrea...l=1#post743605

I take it the oscillations in TPS are from your foot because that car takes a little bit to settle on the dyno? If you could figure out how to do that a little smoother you'd get a lot more data with either filter.

[GHuggins]

That TPS filter is from Cringer's set of filters for DynAir tuning. https://forum.hptuners.com/showthrea...l=1#post743605

I take it the oscillations in TPS are from your foot because that car takes a little bit to settle on the dyno? If you could figure out how to do that a little smoother you'd get a lot more data with either filter.

These are remote tunes collecting data from customers so what most tuners will see. I can say it's too sensitive still for transitioning airflow. Applied it to a whipple 4th gen 5.3 today and it didn't show any of the WOT stuff. Guess it requires several passes to be considered steady state? I couldn't get cringers filters to work at all but I think I need to add some channels before it will.

[SiriusC1024]

Was TPS above the 95% threshold at WOT? Idk why it wouldn't show. Once that is triggered it isn't selective at all and will pass all data.